Re: Restoring invisible and abandoned trials: a call for people to publish the findings
Call to action: RIAT restoration of two fluoxetine studies of depression in children and adolescents
This is a call to action to restore the reporting of the two pivotal placebo-controlled trials of fluoxetine for depression in children and adolescents (1,2). The clinical study reports of these two trials were submitted by Eli Lilly to drug regulators to obtain marketing authorization.
As these trials were misreported, I intend to restore the written record for them in accordance with the principles of the Restoring Invisible and Abandoned Trials (RIAT) initiative (3). On PubMed, there are no links to letters to the editor related to the two trials.
My rationale for correcting the record
I have copies of the two clinical study reports obtained from the Medicines and Healthcare products Regulatory Agency in the UK: B1Y-MC-HCJE and B1Y-MC-X065 (2549 and 1008 pages, respectively). There are very few redactions in these reports, which are mostly limited to names of people and laboratories. Although the authors of the first publication (1) stated that they undertook the study “to evaluate the comparative efficacy, safety, and tolerability of fluoxetine treatment compared with placebo,” there was virtually nothing in their paper about safety. They used two elaborate checking lists for adverse effects but only reported that, “Side effects, as a reason for discontinuation, were minimal, affecting only 4 patients who were receiving fluoxetine and 1 who was receiving placebo. The side effects leading to discontinuation of fluoxetine treatment were in 3 patients in whom manic symptoms developed and 1 patient who developed a severe rash.” The fact that two of the 48 patients on fluoxetine attempted suicide versus none of 48 on placebo was not mentioned.
In the second trial, 109 versus 110 patients were randomised to fluoxetine and placebo, respectively (2). Headache was the only non-solicited adverse event reported significantly more often on fluoxetine than on placebo. However, the clinical study report tells a different story to the published article. As p-values were calculated for each adverse event separately, these analyses lacked appropriate power. I calculated that treatment-related non-solicited adverse events occurred significantly more often on fluoxetine than on placebo (p = 0.02) and, although absolute numbers were small, several events that predispose to suicide and violence (4) were more common on fluoxetine than on placebo (e.g. abnormal dreams, akathisia, hyperkinesia, delusions, emotional lability and hallucinations). The only data in the article (2) about suicidal events or precursor events to suicide and violence were those that led to treatment discontinuation.
Scope and objectives of the restoration
The restoration will focus on adequate reporting of adverse events, in particular those related to suicide and violence; relevant limitations of the trial that were not reported such as issues with the randomisation and blinding procedures; and missing outcomes, which were common, both for effect and adverse effect measures. The restoration will also compare investigator assessed outcomes with those assessed by the patients.
1 Emslie GJ, Rush AJ, Weinberg WA, Kowatch RA, Hughes CW, Carmody T, Rintelmann J. A double-blind, randomized, placebo-controlled trial of fluoxetine in children and adolescents with depression. Arch Gen Psychiatry 1997;54:1031-7.
2 Emslie GJ, Heiligenstein JH, Wagner KD, Hoog SL, Ernest DE, Brown E, Nilsson M, Jacobson JG. Fluoxetine for acute treatment of depression in children and adolescents: a placebo-controlled, randomized clinical trial. J Am Acad Child Adolesc Psychiatry 2002;41:1205-15.
3 Doshi P, Dickersin K, Healy D, Vedula SS, Jefferson T. Restoring invisible and abandoned trials: a call for people to publish the findings. BMJ 2013;346:f2865.
4 Bielefeldt AØ, Danborg PB, Gøtzsche PC. Precursors to suicidality and violence on antidepressants: systematic review of trials in adult healthy volunteers. J R Soc Med 2016;109:381-92.
Rapid Response:
Re: Restoring invisible and abandoned trials: a call for people to publish the findings
Call to action: RIAT restoration of two fluoxetine studies of depression in children and adolescents
This is a call to action to restore the reporting of the two pivotal placebo-controlled trials of fluoxetine for depression in children and adolescents (1,2). The clinical study reports of these two trials were submitted by Eli Lilly to drug regulators to obtain marketing authorization.
As these trials were misreported, I intend to restore the written record for them in accordance with the principles of the Restoring Invisible and Abandoned Trials (RIAT) initiative (3). On PubMed, there are no links to letters to the editor related to the two trials.
My rationale for correcting the record
I have copies of the two clinical study reports obtained from the Medicines and Healthcare products Regulatory Agency in the UK: B1Y-MC-HCJE and B1Y-MC-X065 (2549 and 1008 pages, respectively). There are very few redactions in these reports, which are mostly limited to names of people and laboratories. Although the authors of the first publication (1) stated that they undertook the study “to evaluate the comparative efficacy, safety, and tolerability of fluoxetine treatment compared with placebo,” there was virtually nothing in their paper about safety. They used two elaborate checking lists for adverse effects but only reported that, “Side effects, as a reason for discontinuation, were minimal, affecting only 4 patients who were receiving fluoxetine and 1 who was receiving placebo. The side effects leading to discontinuation of fluoxetine treatment were in 3 patients in whom manic symptoms developed and 1 patient who developed a severe rash.” The fact that two of the 48 patients on fluoxetine attempted suicide versus none of 48 on placebo was not mentioned.
In the second trial, 109 versus 110 patients were randomised to fluoxetine and placebo, respectively (2). Headache was the only non-solicited adverse event reported significantly more often on fluoxetine than on placebo. However, the clinical study report tells a different story to the published article. As p-values were calculated for each adverse event separately, these analyses lacked appropriate power. I calculated that treatment-related non-solicited adverse events occurred significantly more often on fluoxetine than on placebo (p = 0.02) and, although absolute numbers were small, several events that predispose to suicide and violence (4) were more common on fluoxetine than on placebo (e.g. abnormal dreams, akathisia, hyperkinesia, delusions, emotional lability and hallucinations). The only data in the article (2) about suicidal events or precursor events to suicide and violence were those that led to treatment discontinuation.
Scope and objectives of the restoration
The restoration will focus on adequate reporting of adverse events, in particular those related to suicide and violence; relevant limitations of the trial that were not reported such as issues with the randomisation and blinding procedures; and missing outcomes, which were common, both for effect and adverse effect measures. The restoration will also compare investigator assessed outcomes with those assessed by the patients.
1 Emslie GJ, Rush AJ, Weinberg WA, Kowatch RA, Hughes CW, Carmody T, Rintelmann J. A double-blind, randomized, placebo-controlled trial of fluoxetine in children and adolescents with depression. Arch Gen Psychiatry 1997;54:1031-7.
2 Emslie GJ, Heiligenstein JH, Wagner KD, Hoog SL, Ernest DE, Brown E, Nilsson M, Jacobson JG. Fluoxetine for acute treatment of depression in children and adolescents: a placebo-controlled, randomized clinical trial. J Am Acad Child Adolesc Psychiatry 2002;41:1205-15.
3 Doshi P, Dickersin K, Healy D, Vedula SS, Jefferson T. Restoring invisible and abandoned trials: a call for people to publish the findings. BMJ 2013;346:f2865.
4 Bielefeldt AØ, Danborg PB, Gøtzsche PC. Precursors to suicidality and violence on antidepressants: systematic review of trials in adult healthy volunteers. J R Soc Med 2016;109:381-92.
Competing interests: No competing interests